52 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Temperatures of the point X' = 88.7° 

 T> = 67.8° 

 « " Z> = 49.0° 



" " Z = 49.4° 



« « F = 33.2° 



« " X = 12.0° 



As the junctions X, T, etc., were not very accurately placed on their re- 

 spective bars, it would have been desirable to make a corresponding set 

 of observations on these junctions with the X end hot and the X' end 

 cold, if there had been any need of using these last guard-ring tempera- 

 tures in the calculation of the Thomson effect. That there was no such 

 need is shown by the considerations given on p. 35. 



Temperature Coefficient of Thermal Conductivity. 



The measurements designed especially for the determination of this 

 coefficient (see p. 33) were made in February, 1905, after "the observations 

 on the Thomson effect were ended. These especial measurements in- 

 volved few, if any, considerations that have not been set forth in the 

 preceding pages. 



With no electric current in « and /3, but with a permanent gradient of 

 temperature established therein, let 



g h = gradient of temperature at the hot end, 



g c = " « « " " cold " , 



t h = temperature of mid point of hot end couple, 



t c = " " " " " cold " " , 



k h = thermal conductivity at temperature t h , 



k c = " " " " t c , 



f — temperature coefficient of thermal conductivity. 



Then, if there were no lateral flow of heat from a and p 1 , we should 

 have, neglecting the small effect of expansion of the iron, 



h X g k = h c X g c , or ~ = 



h _ ffc 



9h 



(8) /=^F^-(«.-0 = Cr- 1 



(|-i).(,-0. 



